ETS epithelial (ELF/EHF)
ELF1, ELF2, ELF3, ELF4, ELF5 and EHF are the ELF branch of the ETS family. They bind the same GGA(A/T) core as every other ETS protein through the same winged helix-turn-helix domain, but prefer a C where class I ETS factors prefer an A three bases 5' of the core, giving the ten-base site ACCCGGAAGT. ELF3, ELF5 and EHF are the epithelium-specific "ESE" members, expressed in gut, salivary gland, breast, airway, cornea, prostate and bladder epithelium; ELF1, ELF2 and ELF4 are broader, with strong hematopoietic expression. These factors set and maintain epithelial differentiation - mammary alveolar cell fate, intestinal goblet cells - and in airway epithelium EHF represses CFTR; they are recurrently mutated or silenced in carcinomas of the bladder, biliary tract, cervix and breast.
The factors
Six human genes make up this group: ELF1, ELF2 (NERF), ELF3 (ESE-1/ESX), ELF4 (MEF), ELF5 (ESE-2) and EHF (ESE-3). Each has a single ETS domain of about 85 residues - three alpha-helices on a four-stranded antiparallel beta-sheet, a winged helix-turn-helix - that binds DNA as a monomer, with no obligate dimerization partner of the kind GABPA requires 8602247,21548782. Binding profiles of all human and mouse ETS proteins place the ELF proteins in class II (subclass IIa), separate from the large class I group and from ETV6/ETV7 20517297. Expression splits the group in two. ELF3, ELF5 and EHF, together with SPDEF, form the epithelium-specific ETS ("ESE") subfamily: ELF3 is highest in gastrointestinal tract, salivary gland, bladder, liver and prostate; ELF5 is restricted to salivary gland, breast and bladder with some lung, kidney and skin; EHF covers salivary gland, esophagus, prostate, colon, skin, bladder and breast 30200227. ELF1, ELF2 and ELF4 are more widely expressed, with prominent hematopoietic and lymphoid expression.
The motif
The literature writes the class II site as CCCGGAAGT, differing from the class I consensus ACCGGAAGT at a single position: the base three steps 5' of the GGA core, where class I strongly disfavors C and the ELF group accepts or prefers it 20517297,21548782. With the extra 5' adenine that most in vitro models here include, the site reads ACCCGGAAGT, which is what the motifs assigned to this family agree on. The recognition chemistry is identical to the rest of the family: two invariant arginines in helix H3 hydrogen-bond the two guanines of GGA, a wing lysine contacts DNA 5' of the core and a turn lysine contacts the opposite strand 3' of it 8602247, so the core is read directly and the flanks through backbone geometry 21548782. That makes the discriminating information one or two low-content positions, and the six ELF/EHF proteins are not distinguishable from each other by sequence: the JASPAR models for EHF (MA0598.4) and ELF3 (MA0640.3) have the identical consensus CACTTCCTG. One variant is worth recognizing. CCGGAAGT contains a CpG, and methylating that CpG blocks ETS binding 21548782,28473536; the methylation-aware HOCOMOCO models for ELF2, ELF3 and ELF4 in this family instead read CAGGAAGT, a form with no CpG. Whether that is cause or coincidence, both forms are in wide use and a scan should allow CAGGAAG as well as CCGGAAG.
Where it acts
Two contexts dominate. ELF1, ELF2 and ELF4 contribute to the redundant ETS occupancy of CpG-island promoters, where a strong consensus 20-40 bp upstream of the transcription start site is bound interchangeably by many ETS proteins 20517297,21548782; a hit there is a poor basis for naming a factor. The ESE members act mostly at epithelial enhancers and tissue-specific promoters, where their sites are frequently non-consensus and sit among other epithelial regulators. Documented examples include EHF in bronchial epithelium, where it is induced by inflammatory mediators and represses CFTR from an airway-selective element 35 kb upstream of the CFTR promoter 29572268,30200227; EHF across enhancers that define corneal epithelial identity 24142692; and ELF5 at genes of the mammary alveolar program, including repression of SNAI2 18316476,23086238. Because the motif is short and its family-discriminating positions weak, cell type - which ELF genes are actually expressed - is usually more informative about which protein is bound than the sequence is.
Biology and disease
The ESE factors maintain epithelial differentiation. Elf5-null mouse embryos die by about E7.5 from a trophoblast defect 15650748,30200227, and loss of a single Elf5 allele blocks mammary alveolar development in pregnancy 15650748,18316476; ELF5 also holds cells in an epithelial state by repressing SNAI2, and its loss promotes epithelial-mesenchymal transition and metastasis in breast cancer models 23086238. About 30% of Elf3-null embryos die around E11.5 and survivors have abnormal small intestinal epithelium with reduced goblet cells 30200227. The consequences in human cancer are mostly loss of function: ELF3 is recurrently mutated in roughly 6% of bladder cancers, mucinous ovarian cancers, cholangiocarcinomas and gallbladder cancers, about 10% of ampullary carcinomas and about 13% of cervical adenocarcinomas, with a large share of inactivating alleles 30200227; EHF is downregulated with promoter methylation in prostate cancer 30200227. Ehf-null mice are born at Mendelian ratios but develop skin papillomas, corneal ulcers and heightened colitis susceptibility, and gut-specific deletion impairs goblet cell differentiation 34180969.
Reading a hit
A match identifies an ELF/EHF-type ETS site, and it does so only weakly. The sequence separating this family from class I ETS is a single flanking base, so a hit that scores well for ELF3 will almost always also score well for ETS1, ERG or ELK1, and the reverse; if you need to know which protein is bound, use expression and ChIP data, not the motif. The GGAA core is expected millions of times per strand in a 3.1-Gb genome, so context - accessibility, conservation, an epithelial enhancer, neighboring sites - has to carry the inference. Evidence quality is uneven within the family: ELF1, ELF2, ELF3, ELF4 and ELF5 all have in vitro (HT-SELEX or methyl-HT-SELEX) HOCOMOCO models, but EHF's only HOCOMOCO model here is ChIP-derived, so "EHF binds this sequence" rests on weaker evidence than for its relatives. Two of the three Vierstra archetypes grouped here (AC0452:ELF, AC0645:ELF) 32728250 have GGAT-containing consensus sequences and match motifs outside this family better than motifs inside it; treat them as loosely attached. Consensus strings also appear on either strand - CACTTCCTG and CAGGAAGTG are the same site.
How this was worked out
Elf5-null embryos die before E7.5, and loss of one Elf5 allele arrests mammary alveolar development in pregnancy through an epithelial-autonomous defect; a second knockout line confirmed the embryonic lethality and traced it to loss of extraembryonic ectoderm.
Still open then: Which Elf5 target genes, and which DNA sites, carry the mammary phenotype was unknown.
- 2008highKnockout, inducible transgenic overexpression, flow cytometry of mammary progenitors18316476
Elf5 specifies the secretory alveolar lineage: knockout prevents formation of secretory epithelium, inducible overexpression drives alveolar differentiation and milk secretion in virgin mice, and CD61+ luminal progenitors accumulate without it.
- 2010highHigh-throughput microwell binding assay, protein-binding microarrays, site-directed mutagenesis, ChIP-seq20517297
Profiling all human and mouse ETS proteins puts the ELF proteins in class II, which tolerates the C three bases 5' of the core that class I rejects, a difference traced to a single leucine where class I carries phenylalanine or tyrosine; ETV6 and ETV7 are split off as class IIb, and ELF1 ChIP-seq peaks are strongly promoter-enriched.
Still open then: No sequence feature separated one class IIa ELF protein from another.
- 2012mediumConditional knockout, EMT and metastasis models, MMTV-neu tumour model, clinical breast cancer samples23086238
Elf5 suppresses the epithelial-mesenchymal transition in mammary development and breast cancer metastasis by directly repressing Snail2.
EHF promotes corneal epithelial fate through both activating and repressing target genes, with likely interplay with KLF4 and KLF5.
Methylation-sensitive SELEX across 542 human transcription factors finds ETS-family proteins generally inhibited by CpG methylation, which bears directly on the CpG inside CCGGAAGT.
Still open then: How methylation shapes which ELF sites are used in cells was not measured.
- 2018mediumChIP-seq, RNA-seq after depletion, siRNA screen, ChIP in primary bronchial epithelial cells28461336 29572268
In bronchial epithelium EHF targets inflammation and wound-repair genes, and an siRNA screen of about 1500 factors identifies it, with KLF5, as one of the strongest repressors of CFTR, acting at an airway-selective element 35 kb upstream of the CFTR promoter.
Still open then: How the cystic fibrosis modifier variant near EHF changes EHF activity was not established.
Ehf-null mice are born at Mendelian ratios but develop facial skin papillomas, preputial or vulval abscesses and corneal ulcers and are more susceptible to colitis; gut-specific deletion impairs goblet cell differentiation and enhances Apc-initiated adenomas. This overturns the 2018 review statement that no Ehf deletion phenotype had been reported.
Still unresolved
- Why the methylation-aware HOCOMOCO models for ELF2, ELF3 and ELF4 read CAGGAAGT while the unmethylated in vitro models read CCGGAAGT is not established: the A may simply avoid a CpG whose methylation blocks ETS binding, or it may be a genuine second preference that persists in cells.
- Two footprint archetypes grouped here read GGAT, whereas the ESE literature describes ELF3, ELF5 and EHF as GGAA binders and class II proteins as more sensitive than class I to the A-to-T change that creates GGAT; whether these archetypes record real ELF binding at GGAT sites or SPDEF-like footprints is unresolved.
- Whether ELF3 acts as a tumour suppressor or a tumour promoter depends on context and is unsettled: inactivating, often heterozygous, mutations in bladder, ovarian and biliary cancers argue for loss of function, while amplification in colorectal and gastric cancer and raised expression in prostate cancer and mutant cervical adenocarcinomas point the other way.
The one base that distinguishes this family from class I ETS is a low-information flanking position, so in practice a hit means "an ETS site" and not "an ELF site", let alone a specific ELF protein; the JASPAR EHF and ELF3 models are literally the same consensus. Which of the six proteins is bound has to come from expression or ChIP, not from sequence. Within the family, EHF's only HOCOMOCO model here is ChIP-derived rather than in vitro, so its label carries less weight than ELF1-ELF5, which have HT-SELEX or methyl-HT-SELEX support. Two of the three Vierstra ELF archetypes in this family have GGAT-type consensus sequences and their nearest neighbours lie outside the family. Allow both CCGGAAG and the CpG-free CAGGAAG form when scanning.
References
- Wei GH, Badis G, Berger MF, et al. Genome-wide analysis of ETS-family DNA-binding in vitro and in vivo. EMBO J 2010;29:2147-60. PMID 20517297
- Hollenhorst PC, McIntosh LP, Graves BJ. Genomic and biochemical insights into the specificity of ETS transcription factors. Annu Rev Biochem 2011;80:437-71. PMID 21548782
- Luk IY, Reehorst CM, Mariadason JM. ELF3, ELF5, EHF and SPDEF Transcription Factors in Tissue Homeostasis and Cancer. Molecules 2018;23. PMID 30200227
- Kodandapani R, Pio F, Ni CZ, et al. A new pattern for helix-turn-helix recognition revealed by the PU.1 ETS-domain-DNA complex. Nature 1996;380:456-60. PMID 8602247
- Yin Y, Morgunova E, Jolma A, et al. Impact of cytosine methylation on DNA binding specificities of human transcription factors. Science 2017;356. PMID 28473536
- Oakes SR, Naylor MJ, Asselin-Labat ML, et al. The Ets transcription factor Elf5 specifies mammary alveolar cell fate. Genes Dev 2008;22:581-6. PMID 18316476
- Chakrabarti R, Hwang J, Andres Blanco M, et al. Elf5 inhibits the epithelial-mesenchymal transition in mammary gland development and breast cancer metastasis by transcriptionally repressing Snail2. Nat Cell Biol 2012;14:1212-22. PMID 23086238
- Fossum SL, Mutolo MJ, Tugores A, et al. Ets homologous factor (EHF) has critical roles in epithelial dysfunction in airway disease. J Biol Chem 2017;292:10938-10949. PMID 28461336
- Stephens DN, Klein RH, Salmans ML, et al. The Ets transcription factor EHF as a regulator of cornea epithelial cell identity. J Biol Chem 2013;288:34304-24. PMID 24142692
- Vierstra J, Lazar J, Sandstrom R, et al. Global reference mapping of human transcription factor footprints. Nature 2020;583:729-736. PMID 32728250
- Zhou J, Chehab R, Tkalcevic J, et al. Elf5 is essential for early embryogenesis and mammary gland development during pregnancy and lactation. EMBO J 2005;24:635-44. PMID 15650748
- Mutolo MJ, Leir SH, Fossum SL, et al. A transcription factor network represses CFTR gene expression in airway epithelial cells. Biochem J 2018;475:1323-1334. PMID 29572268
- Reehorst CM, Nightingale R, Luk IY, et al. EHF is essential for epidermal and colonic epithelial homeostasis, and suppresses Apc-initiated colonic tumorigenesis. Development 2021;148. PMID 34180969
25 motifs in this family — JASPAR 6 · HOCOMOCO 10 · CIS-BP 6 · Vierstra 3
| motif | database | consensus | bits |
|---|---|---|---|
| AC0452:ELF:Ets | Vierstra | TGTGCGGATGC | 18.4 |
| ELF5.H14CORE.1.S.BSB | HOCOMOCO | AcCCGGAAGT | 17.2 |
| M03002_2.00 | CIS-BP | ACCCGGAAGTa | 17 |
| ELF4 (MA0641.1) | JASPAR | ACCCGGAAGT | 16.3 |
| M02966_2.00 | CIS-BP | ACCCGGAAGT | 16.3 |
| AC0645:ELF:Ets | Vierstra | ATaaCGGATG | 15.5 |
| ELF3.H14CORE.1.PM.APMA | HOCOMOCO | AcCAGGAAGTa | 15.1 |
| ELF4.H14CORE.1.M.BMB | HOCOMOCO | AnCAGGAAGTa | 14.9 |
| M07947_2.00 | CIS-BP | acCCGGAAGtG | 14 |
| EHF.H14CORE.0.P.BPB | HOCOMOCO | AccAGGAAGT | 13.9 |
| ELF2.H14CORE.1.M.BMB | HOCOMOCO | anCaGGAAGTa | 13.8 |
| ELF3.H14CORE.0.S.BSB | HOCOMOCO | AcCCGGAAGT | 13.3 |
| ELF4.H14CORE.0.PS.APSA | HOCOMOCO | acCCGGAAGTG | 13.3 |
| ELF1.H14CORE.0.PSM.APSMA | HOCOMOCO | CCGGAAGTg | 13.1 |
| M02733_2.00 | CIS-BP | ACCCGGAAGT | 13.1 |
| ELF2.H14CORE.0.PS.APSA | HOCOMOCO | CCGGAAGTG | 13 |
| M04722_2.00 | CIS-BP | accCGGAAGT | 13 |
| ELF2 (MA1483.3) | JASPAR | acCCGGAAGT | 12.9 |
| ELF5.H14CORE.0.PSM.APSMA | HOCOMOCO | AnvaGGAAGT | 12.5 |
| ELF1 (MA0473.4) | JASPAR | CAGGAAGTG | 12.4 |
| M02988_2.00 | CIS-BP | AccCGGAAGT | 11.6 |
| EHF (MA0598.4) | JASPAR | cACTTCCTg | 11.4 |
| ELF3 (MA0640.3) | JASPAR | cACTTCCTg | 11.3 |
| Elf5 (MA0136.4) | JASPAR | AAGGAAGT | 11.1 |
| AC0640:ELF/ETV:Ets | Vierstra | CCGGAAGT | 10.3 |
HOCOMOCO tags read as assay + quality grade: P ChIP-seq, S HT-SELEX, M methyl-HT-SELEX, G GHT-SELEX, I SMiLE-seq, B PBM; grade A means two or more assay types agreed. A P-only matrix is a record of where the protein was crosslinked, which need not be a sequence it binds itself.
The factors, elsewhere
These entries are written from the primary literature and every citation is checked against PubMed, but this one can still be out of date, too confident, or simply wrong — especially where a motif’s published site and its database matrices disagree. Corrections and additions are welcome, and they are the fastest way to make this page better for everyone.